Annotation of parser3/src/main/pa_common.C, revision 1.100
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.100 ! paf 4: Copyright(c) 2001, 2002 ArtLebedev Group(http://www.artlebedev.com)
1.85 paf 5: Author: Alexander Petrosyan <paf@design.ru>(http://paf.design.ru)
1.16 paf 6:
1.100 ! paf 7: $Id: pa_common.C,v 1.99 2002/02/07 17:41:36 paf Exp $
1.1 paf 8: */
9:
10: #include "pa_common.h"
1.2 paf 11: #include "pa_types.h"
1.4 paf 12: #include "pa_exception.h"
1.14 paf 13: #include "pa_pool.h"
14: #include "pa_globals.h"
15: #include "pa_value.h"
16: #include "pa_hash.h"
17: #include "pa_string.h"
1.1 paf 18:
1.98 paf 19: #ifdef WIN32
20: # include <windows.h>
21: #endif
22:
1.93 paf 23: // some maybe-undefined constants
24:
1.82 paf 25: #ifndef _O_TEXT
26: # define _O_TEXT 0
27: #endif
28: #ifndef _O_BINARY
29: # define _O_BINARY 0
1.47 paf 30: #endif
1.80 paf 31: #ifndef O_TRUNC
32: # define O_TRUNC 0
33: #endif
34:
1.93 paf 35: // locking constants
36:
1.99 paf 37: #ifdef HAVE_FLOCK
38:
39: static int lock_shared_blocking(int fd) { return flock(fd, LOCK_SH); }
40: static int lock_exclusive_blocking(int fd) { return flock(fd, LOCK_EX); }
41: static int lock_exclusive_nonblocking(int fd) { return flock(fd, LOCK_EX || LOCK_NB); }
42: static int unlock(int fd) { return flock(fd, LOCK_UN); }
43:
1.98 paf 44: #else
1.99 paf 45: #ifdef HAVE__LOCKING
1.98 paf 46:
1.99 paf 47: #define FLOCK(operation) lseek(fd, 0, SEEK_SET); return _locking(fd, operation, 1)
48: static int lock_shared_blocking(int fd) { FLOCK(_LK_LOCK); }
49: static int lock_exclusive_blocking(int fd) { FLOCK(_LK_LOCK); }
50: static int lock_exclusive_nonblocking(int fd) { FLOCK(_LK_NBLCK); }
51: static int unlock(int fd) { FLOCK(_LK_UNLCK); }
1.93 paf 52:
1.99 paf 53: #else
54: #ifdef HAVE_FCNTL
1.93 paf 55:
1.99 paf 56: #define FLOCK(cmd, arg) struct flock ls={arg, SEEK_SET}; return fcntl(fd, cmd, &ls)
57: static int lock_shared_blocking(int fd) { FLOCK(F_SETLKW, F_RDLCK); }
58: static int lock_exclusive_blocking(int fd) { FLOCK(F_SETLKW, F_WRLCK); }
59: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_SETLK, F_RDLCK); }
60: static int unlock(int fd) { FLOCK(F_SETLK, F_UNLCK); }
1.93 paf 61:
62: #else
63: #ifdef HAVE_LOCKF
1.99 paf 64:
65: #define FLOCK(fd, operation) lseek(fd, 0, SEEK_SET); return lockf(fd, operation, 1)
66: static int lock_shared_blocking(int fd) { FLOCK(F_LOCK); } // on intel solaris man doesn't have doc on shared blocking
67: static int lock_exclusive_blocking(int fd) { FLOCK(F_LOCK); }
68: static int lock_exclusive_nonblocking(int fd) { FLOCK(F_TLOCK); }
69: static int unlock(int fd) { FLOCK(F_TLOCK); }
70:
1.93 paf 71: #else
1.99 paf 72:
73: #error unable to find file locking func
74:
75: #endif
1.93 paf 76: #endif
77: #endif
78: #endif
79:
1.86 paf 80: static char *strnchr(char *buf, size_t size, char c) {
81: for(; size-->0; buf++) {
82: if(*buf==c)
83: return buf;
84: }
85:
86: return 0;
87: }
88:
89: void fix_line_breaks(char *buf, size_t& size) {
1.87 paf 90: //_asm int 3;
1.86 paf 91: const char * const eob=buf+size;
1.87 paf 92: char *dest=buf;
1.72 parser 93: // fix DOS: \r\n -> \n
94: // fix Macintosh: \r -> \n
1.87 paf 95: char *bol=buf;
1.86 paf 96: while(char *eol=strnchr(bol, eob -bol, '\r')) {
1.72 parser 97: size_t len=eol-bol;
98: if(dest!=bol)
99: memcpy(dest, bol, len);
100: dest+=len;
101: *dest++='\n';
102:
1.87 paf 103: if(&eol[1]<eob && eol[1]=='\n') { // \r,\n = DOS
1.72 parser 104: bol=eol+2;
105: size--;
106: } else // \r,not \n = Macintosh
107: bol=eol+1;
108: }
109: // last piece without \r, including terminating 0
110: if(dest!=bol)
1.87 paf 111: memcpy(dest, bol, eob-bol);
1.72 parser 112: }
1.18 paf 113:
1.28 paf 114: char *file_read_text(Pool& pool, const String& file_spec, bool fail_on_read_problem) {
1.72 parser 115: void *result; size_t size;
116: return file_read(pool, file_spec, result, size, true, fail_on_read_problem)?(char *)result:0;
1.34 paf 117: }
118: bool file_read(Pool& pool, const String& file_spec,
1.87 paf 119: void*& data, size_t& data_size, bool as_text,
1.44 paf 120: bool fail_on_read_problem,
121: size_t offset, size_t limit) {
1.64 parser 122: const char *fname=file_spec.cstr(String::UL_FILE_SPEC);
1.62 parser 123: //printf("file_read(%s)\n", fname);
1.33 paf 124: int f;
1.2 paf 125: struct stat finfo;
1.33 paf 126:
127: // first open, next stat:
1.45 paf 128: // directory update of NTFS hard links performed on open.
1.33 paf 129: // ex:
130: // a.html:^test[] and b.html hardlink to a.html
131: // user inserts ! before ^test in a.html
132: // directory entry of b.html in NTFS not updated at once,
1.35 paf 133: // they delay update till open, so we would receive "!^test[" string
134: // if would do stat, next open.
1.98 paf 135: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.99 paf 136: lock_shared_blocking(f);
1.98 paf 137: if(stat(fname, &finfo)!=0) {
138: Exception e(0, 0,
139: &file_spec,
140: "stat failed: %s (%d), actual filename '%s'",
141: strerror(errno), errno, fname);
1.99 paf 142: unlock(f);
1.98 paf 143: close(f);
144: if(fail_on_read_problem)
145: throw e;
146: return false;
147: }
148: size_t max_size=limit?min(offset+limit, (size_t)finfo.st_size)-offset:finfo.st_size;
1.47 paf 149: if(!max_size) { // eof
1.83 paf 150: if(as_text) {
151: data=pool.malloc(1);
152: *(char*)data=0;
153: } else
154: data=0;
1.98 paf 155: data_size=0;
1.59 parser 156: } else {
1.78 paf 157: data=pool.malloc(max_size+(as_text?1:0), 3);
1.59 parser 158: if(offset)
159: lseek(f, offset, SEEK_SET);
1.98 paf 160: data_size=read(f, data, max_size);
1.47 paf 161: }
1.99 paf 162: unlock(f);
1.2 paf 163: close(f);
1.59 parser 164: if(!max_size) // eof
165: return true;
1.32 paf 166:
1.98 paf 167: if(int(data_size)<0 || data_size>max_size)
1.73 parser 168: throw Exception(0, 0,
1.33 paf 169: &file_spec,
1.44 paf 170: "read failed: actually read %d bytes count not in [0..%lu] valid range",
1.98 paf 171: data_size, (unsigned long)max_size); //never
1.87 paf 172:
173: if(as_text) {
174: fix_line_breaks((char *)data, data_size);
175: // note: after fixing
176: ((char*&)data)[data_size]=0;
177: }
1.72 parser 178: return true;
1.2 paf 179: }
1.4 paf 180: if(fail_on_read_problem)
1.73 parser 181: throw Exception(0, 0,
1.33 paf 182: &file_spec,
1.54 parser 183: "read failed: %s (%d), actual filename '%s'",
184: strerror(errno), errno, fname);
1.34 paf 185: return false;
1.8 paf 186: }
187:
1.63 parser 188: static void create_dir_for_file(const String& file_spec) {
189: size_t pos_after=1;
190: int pos_before;
191: while((pos_before=file_spec.pos("/", 1, pos_after))>=0) {
1.64 parser 192: mkdir(file_spec.mid(0, pos_before).cstr(String::UL_FILE_SPEC), 0775);
1.63 parser 193: pos_after=pos_before+1;
194: }
195: }
196:
1.98 paf 197: bool file_write_action_under_lock(
1.28 paf 198: const String& file_spec,
1.96 paf 199: const char *action_name, void (*action)(int, void *), void *context,
1.80 paf 200: bool as_text,
1.98 paf 201: bool do_append,
202: bool do_block) {
1.64 parser 203: const char *fname=file_spec.cstr(String::UL_FILE_SPEC);
1.28 paf 204: int f;
1.80 paf 205: if(access(fname, W_OK)!=0) // no
1.63 parser 206: create_dir_for_file(file_spec);
1.50 paf 207:
1.80 paf 208: if((f=open(fname,
209: O_CREAT|O_RDWR
210: |(as_text?_O_TEXT:_O_BINARY)
1.98 paf 211: |(do_append?O_APPEND:0), 0664))>=0) {
1.99 paf 212: if((do_block?lock_exclusive_blocking(f):lock_exclusive_nonblocking(f))!=0) {
1.98 paf 213: close(f);
214: return false;
215: }
1.96 paf 216:
217: try {
218: action(f, context);
219: } catch(...) {
1.98 paf 220: #if O_TRUNC==0
221: ftruncate(f, tell(f));
222: #endif
1.99 paf 223: unlock(f);
1.96 paf 224: close(f);
225: /*re*/throw;
226: }
1.80 paf 227:
1.98 paf 228: #if O_TRUNC==0
229: ftruncate(f, tell(f));
230: #endif
1.99 paf 231: unlock(f);
1.80 paf 232: close(f);
1.98 paf 233: return true;
1.80 paf 234: } else
235: throw Exception(0, 0,
236: &file_spec,
1.96 paf 237: "%s failed: %s (%d), actual filename '%s'",
238: action_name, strerror(errno), errno, fname);
239: // here should be nothing, see rethrow above
240: }
241:
242: #ifndef DOXYGEN
243: struct File_write_action_info {
244: const void *data; size_t size;
245: };
246: #endif
247: static void file_write_action(int f, void *context) {
248: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
249: if(info.size)
250: write(f, info.data, info.size);
251: }
252: void file_write(
253: const String& file_spec,
254: const void *data, size_t size,
255: bool as_text,
256: bool do_append) {
257: File_write_action_info info={data, size};
1.98 paf 258: file_write_action_under_lock(
1.96 paf 259: file_spec,
260: "write", file_write_action, &info,
261: as_text,
262: do_append);
1.30 paf 263: }
264:
1.63 parser 265: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 266: static void rmdir(const String& file_spec, size_t pos_after) {
267: int pos_before;
268: if((pos_before=file_spec.pos("/", 1, pos_after))>=0)
269: rmdir(file_spec, pos_before+1);
270:
1.64 parser 271: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::UL_FILE_SPEC));
1.50 paf 272: }
1.95 paf 273: bool file_delete(const String& file_spec, bool fail_on_read_problem) {
1.64 parser 274: const char *fname=file_spec.cstr(String::UL_FILE_SPEC);
1.54 parser 275: if(unlink(fname)!=0)
1.93 paf 276: if(fail_on_read_problem)
277: throw Exception(0, 0,
278: &file_spec,
279: "unlink failed: %s (%d), actual filename '%s'",
280: strerror(errno), errno, fname);
281: else
282: return false;
1.50 paf 283:
284: rmdir(file_spec, 1);
1.93 paf 285: return true;
1.60 parser 286: }
1.95 paf 287: void file_move(const String& old_spec, const String& new_spec) {
1.64 parser 288: const char *old_spec_cstr=old_spec.cstr(String::UL_FILE_SPEC);
289: const char *new_spec_cstr=new_spec.cstr(String::UL_FILE_SPEC);
1.63 parser 290:
291: create_dir_for_file(new_spec);
292:
1.60 parser 293: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.73 parser 294: throw Exception(0, 0,
1.60 parser 295: &old_spec,
296: "rename failed: %s (%d), actual filename '%s' to '%s'",
297: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 298:
299: rmdir(old_spec, 1);
1.31 paf 300: }
301:
1.51 paf 302:
303: static bool entry_readable(const String& file_spec, bool need_dir) {
1.64 parser 304: const char *fname=file_spec.cstr(String::UL_FILE_SPEC);
1.51 paf 305: struct stat finfo;
306: if(access(fname, R_OK)==0 && stat(fname, &finfo)==0) {
1.73 parser 307: bool is_dir=finfo.st_mode&S_IFDIR != 0;
1.51 paf 308: return is_dir==need_dir;
309: }
310: return false;
311: }
1.31 paf 312: bool file_readable(const String& file_spec) {
1.51 paf 313: return entry_readable(file_spec, false);
314: }
315: bool dir_readable(const String& file_spec) {
316: return entry_readable(file_spec, true);
1.65 parser 317: }
318: String *file_readable(const String& path, const String& name) {
319: String *result=new(path.pool()) String(path);
320: *result << "/";
321: *result << name;
322: return file_readable(*result)?result:0;
1.43 paf 323: }
324: bool file_executable(const String& file_spec) {
1.64 parser 325: return access(file_spec.cstr(String::UL_FILE_SPEC), X_OK)==0;
1.44 paf 326: }
327:
1.64 parser 328: bool file_stat(const String& file_spec,
1.58 parser 329: size_t& rsize,
330: time_t& ratime,
331: time_t& rmtime,
1.64 parser 332: time_t& rctime,
333: bool fail_on_read_problem) {
1.44 paf 334: Pool& pool=file_spec.pool();
1.64 parser 335: const char *fname=file_spec.cstr(String::UL_FILE_SPEC);
1.44 paf 336: struct stat finfo;
337: if(stat(fname, &finfo)!=0)
1.64 parser 338: if(fail_on_read_problem)
1.73 parser 339: throw Exception(0, 0,
1.67 parser 340: &file_spec,
341: "getting file size failed: %s (%d), real filename '%s'",
342: strerror(errno), errno, fname);
1.64 parser 343: else
344: return false;
1.58 parser 345: rsize=finfo.st_size;
346: ratime=finfo.st_atime;
347: rmtime=finfo.st_mtime;
348: rctime=finfo.st_ctime;
1.64 parser 349: return true;
1.18 paf 350: }
351:
1.8 paf 352: char *getrow(char **row_ref, char delim) {
353: char *result=*row_ref;
354: if(result) {
355: *row_ref=strchr(result, delim);
356: if(*row_ref)
357: *((*row_ref)++)=0;
358: else if(!*result)
359: return 0;
360: }
361: return result;
362: }
363:
1.23 paf 364: char *lsplit(char *string, char delim) {
365: if(string) {
366: char *v=strchr(string, delim);
1.8 paf 367: if(v) {
368: *v=0;
369: return v+1;
370: }
371: }
372: return 0;
373: }
374:
375: char *lsplit(char **string_ref, char delim) {
376: char *result=*string_ref;
377: char *next=lsplit(*string_ref, delim);
378: *string_ref=next;
379: return result;
1.9 paf 380: }
381:
382: char *rsplit(char *string, char delim) {
1.18 paf 383: if(string) {
1.9 paf 384: char *v=strrchr(string, delim);
1.18 paf 385: if(v) {
1.9 paf 386: *v=0;
387: return v+1;
388: }
389: }
390: return NULL;
1.10 paf 391: }
392:
1.37 paf 393: /// @todo less stupid type detection
1.10 paf 394: char *format(Pool& pool, double value, char *fmt) {
1.78 paf 395: char *result=(char *)pool.malloc(MAX_NUMBER, 4);
1.10 paf 396: if(fmt)
397: if(strpbrk(fmt, "diouxX"))
398: if(strpbrk(fmt, "ouxX"))
1.33 paf 399: snprintf(result, MAX_NUMBER, fmt, (uint)value );
1.10 paf 400: else
1.33 paf 401: snprintf(result, MAX_NUMBER, fmt, (int)value );
1.10 paf 402: else
403: snprintf(result, MAX_NUMBER, fmt, value);
404: else
1.33 paf 405: snprintf(result, MAX_NUMBER, "%d", (int)value);
1.10 paf 406:
407: return result;
1.12 paf 408: }
409:
1.36 paf 410: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 411: #ifdef WIN32
412: do{
413: int chunk_written=fwrite(buf, 1, min(8*0x400, size), stdout);
414: if(chunk_written<=0)
415: break;
416: size-=chunk_written;
1.36 paf 417: buf=((const char*)buf)+chunk_written;
1.12 paf 418: } while(size>0);
419:
420: return size;
421: #else
1.13 paf 422: return fwrite(buf, 1, size, stdout);
1.12 paf 423: #endif
1.2 paf 424: }
1.14 paf 425:
1.77 parser 426: char *unescape_chars(Pool& pool, const char *cp, int len) {
1.78 paf 427: char *s=(char *)pool.malloc(len + 1, 5);
1.14 paf 428: enum EscapeState {
1.33 paf 429: EscapeRest,
430: EscapeFirst,
1.14 paf 431: EscapeSecond
432: } escapeState=EscapeRest;
433: int escapedValue=0;
434: int srcPos=0;
435: int dstPos=0;
436: while(srcPos < len) {
437: int ch=cp[srcPos];
438: switch(escapeState) {
439: case EscapeRest:
440: if(ch=='%') {
441: escapeState=EscapeFirst;
442: } else if(ch=='+') {
443: s[dstPos++]=' ';
444: } else {
445: s[dstPos++]=ch;
446: }
447: break;
448: case EscapeFirst:
449: escapedValue=hex_value[ch] << 4;
450: escapeState=EscapeSecond;
451: break;
452: case EscapeSecond:
453: escapedValue +=hex_value[ch];
454: s[dstPos++]=escapedValue;
455: escapeState=EscapeRest;
456: break;
457: }
458: srcPos++;
459: }
460: s[dstPos]=0;
461: return s;
462: }
463:
1.52 paf 464: /// used by attributed_meaning_to_string / append_attribute_subattribute
1.40 paf 465: struct Attributed_meaning_info {
1.52 paf 466: String *header; // header line being constructed
467: String::Untaint_lang lang; // language in which to append to that line
1.40 paf 468: };
1.17 paf 469: static void append_attribute_subattribute(const Hash::Key& akey, Hash::Val *avalue,
1.14 paf 470: void *info) {
471: if(akey==VALUE_NAME)
472: return;
473:
1.40 paf 474: Attributed_meaning_info& ami=*static_cast<Attributed_meaning_info *>(info);
475:
1.14 paf 476: // ...; charset=windows1251
1.48 paf 477: *ami.header << "; ";
1.70 parser 478: ami.header->append(akey, ami.lang);
1.48 paf 479: *ami.header << "=";
1.70 parser 480: ami.header->append(static_cast<Value *>(avalue)->as_string(), ami.lang);
1.14 paf 481: }
1.49 paf 482: const String& attributed_meaning_to_string(Value& meaning,
483: String::Untaint_lang lang) {
1.20 paf 484: String &result=*new(meaning.pool()) String(meaning.pool());
1.75 parser 485: if(Hash *hash=meaning.get_hash(0)) {
1.14 paf 486: // $value(value) $subattribute(subattribute value)
487: if(Value *value=static_cast<Value *>(hash->get(*value_name)))
1.40 paf 488: result.append(value->as_string(), lang, true);
1.14 paf 489:
1.40 paf 490: Attributed_meaning_info attributed_meaning_info={
1.52 paf 491: &result,
492: lang
1.40 paf 493: };
494: hash->for_each(append_attribute_subattribute, &attributed_meaning_info);
1.14 paf 495: } else // result value
1.40 paf 496: result.append(meaning.as_string(), lang, true);
1.14 paf 497:
498: return result;
1.24 paf 499: }
500:
501: #ifdef WIN32
502: void back_slashes_to_slashes(char *s) {
503: if(s)
504: for(; *s; s++)
505: if(*s=='\\')
506: *s='/';
507: }
1.42 paf 508: /*
509: void slashes_to_back_slashes(char *s) {
510: if(s)
511: for(; *s; s++)
512: if(*s=='/')
513: *s='\\';
514: }
515: */
1.24 paf 516: #endif
1.41 paf 517:
518: bool StrEqNc(const char *s1, const char *s2, bool strict) {
519: while(true) {
520: if(!(*s1)) {
521: if(!(*s2))
522: return true;
523: else
524: return !strict;
525: } else if(!(*s2))
526: return !strict;
527: if(isalpha(*s1)) {
528: if(tolower(*s1) !=tolower(*s2))
529: return false;
530: } else if((*s1) !=(*s2))
531: return false;
532: s1++;
533: s2++;
534: }
1.57 parser 535: }
536:
1.84 paf 537: static bool isLeap(int year) {
1.57 parser 538: return !(
539: (year % 4) || ((year % 400) && !(year % 100))
540: );
541: }
542:
543: int getMonthDays(int year, int month) {
544: int monthDays[]={
545: 31,
546: isLeap(year) ? 29 : 28,
547: 31,
548: 30,
549: 31,
550: 30,
551: 31,
552: 31,
553: 30,
554: 31,
555: 30,
556: 31
557: };
558: return monthDays[month];
1.41 paf 559: }
1.69 parser 560:
561: void remove_crlf(char *start, char *end) {
562: for(char *p=start; p<end; p++)
563: switch(*p) {
564: case '\n': *p='|'; break;
565: case '\r': *p=' '; break;
566: }
1.91 paf 567: }
568:
569:
570: /// must be last in this file
571: #undef vsnprintf
572: int __vsnprintf(char *b, size_t s, const char *f, va_list l) {
573: if(!s)
574: return 0;
575:
576: int r;
577: // note: on win32& maybe somewhere else
578: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
579: --s;
580: #if _MSC_VER
581: /*
582: win32:
583: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
584:
585: if the number of bytes to write exceeds buffer, then count bytes are written and –1 is returned
586: */
587: r=_vsnprintf(b, s, f, l);
588: if(r<0)
589: r=s;
590: #else
591: r=vsnprintf(b, s, f, l);
592: /*
593: solaris:
594: man vsnprintf
595:
596: The snprintf() function returns the number of characters
597: formatted, that is, the number of characters that would have
598: been written to the buffer if it were large enough. If the
599: value of n is 0 on a call to snprintf(), an unspecified
600: value less than 1 is returned.
601: */
602:
603: if(r<0)
604: r=0;
605: else if(r>s)
606: r=s;
607: #endif
608: b[r]=0;
609: return r;
610: }
611:
612: int __snprintf(char *b, size_t s, const char *f, ...) {
613: va_list l;
614: va_start(l, f);
615: int r=__vsnprintf(b, s, f, l);
616: va_end(l);
617: return r;
1.98 paf 618: }
619:
620: int pa_sleep(unsigned long secs, unsigned long usecs) {
621: for (; usecs >= 1000000; ++secs, usecs -= 1000000);
622:
623: #ifdef WIN32
624: Sleep(secs * 1000 + usecs / 1000);
625: return 0;
626: #else
627: struct timeval t;
628: t.tv_sec = secs;
629: t.tv_usec = usecs;
630: return (select(0, NULL, NULL, NULL, &t) == -1 ? errno : 0);
631: #endif
1.74 parser 632: }
E-mail: